Ticket 6095

Summary: Wrong CPU times reported by cgroup accounting plugin
Product: Slurm Reporter: Bob Dröge <b.e.droge>
Component: AccountingAssignee: Jacob Jenson <jacob>
Status: RESOLVED DUPLICATE QA Contact:
Severity: 6 - No support contract    
Priority: --- CC: paddy
Version: 18.08.3   
Hardware: Linux   
OS: Linux   
Site: -Other- Slinky Site: ---
Alineos Sites: --- Atos/Eviden Sites: ---
Confidential Site: --- Coreweave sites: ---
Cray Sites: --- DS9 clusters: ---
Google sites: --- HPCnow Sites: ---
HPE Sites: --- IBM Sites: ---
NOAA SIte: --- NoveTech Sites: ---
Nvidia HWinf-CS Sites: --- OCF Sites: ---
Recursion Pharma Sites: --- SFW Sites: ---
SNIC sites: --- Tzag Elita Sites: ---
Linux Distro: CentOS Machine Name:
CLE Version: Version Fixed:
Target Release: --- DevPrio: ---
Emory-Cloud Sites: ---

Description Bob Dröge 2018-11-27 07:52:29 MST 
Since a recent upgrade from SLURM 17.11.5 to 18.08.3 we're getting weird CPU times reported by the sstat and sacct tools. We use the cgroup plugin for job accounting:
JobAcctGatherType=jobacct_gather/cgroup

For instance, a simple job that uses  a single core for about a minute produces the following TotalCPU record in the accounting database:

$ sacct -j 3030633 -o AllocCPUs,Elapsed,TotalCPU,MinCPU
 AllocCPUS    Elapsed   TotalCPU     MinCPU 
---------- ---------- ---------- ---------- 
         1   00:01:01   01:34:35            
         1   00:01:01   01:34:35   01:34:34 

Similar values appear in all the other CPU time related fiels, also with sstat. It looks like all these values are off by a factor of 100.

In order to reproduce the issue, I just installed a clean/minimal version of SLURM 18.08.3 in a VM and found the same behaviour; with the linux accounting plugin, I do get correct CPU times.
Comment 1 Michael Hinton 2019-01-15 14:42:07 MST 
Hey Bob,

This was fixed today. Check out bug 6332.

Thanks,
Michael

*** This ticket has been marked as a duplicate of ticket 6332 ***
Comment 2 Bob Dröge 2019-01-16 01:33:21 MST 
(In reply to Michael Hinton from comment #1)
> Hey Bob,
> 
> This was fixed today. Check out bug 6332.
> 
> Thanks,
> Michael
> 
> *** This bug has been marked as a duplicate of bug 6332 ***


Great, thank you!